How to call an Oracle procedure and commit from .net app

[tev]

We developed a procedure that performs some DML transactions and commit if the DML is successful or rollback if otherwise. We also developed a .net application to call the procedure and display the return status to the application user. The major challenge we have now is that some calls to the procedure may time out as a result of network connectivity which the application interprete as transaction failure whereas the procedure may still go ahead to complete the transaction and subsequently commit it. The result of
Customer Problem Description
---------------------------------------------------

Problem Summary
---------------------------------------------------
How to call an Oracle procedure and commit/rollback from a .net application

Problem Description
---------------------------------------------------
We developed a procedure that performs some DML transactions and commit if the DML is successful or rollback if otherwise. We also developed a .net application to call the procedure and display the return status to the application user. The major challenge we have now is that some calls to the procedure may time out as a result of network connectivity which the application interprete as transaction failure whereas the procedure may still go ahead to complete the transaction and subsequently commit it. The result of this is that user may end up carrying out the same transaction over and over again based on the failure message displayed by the application. They get to know that all the transactions were successful when they go back to display all the transaction done from the database.
We need to know how to ensure that the .net application take control of the commit and rollback operation after calling the DML. How can we go about this? We will appreciate a sample .net calls snippet

[tev]

Also, in some situation , the procedure may take longer that the time out of the calling application to process and in this situation , we noticed that the procedure may process the transaction while the application may return a failed message to the user. What is your advice ? How can we avoid this scenario

[Tim...]

Hi.

Most client languages, include those supported by .NET allow you to manage transactions. To do this, you need to remove any commit/rollback control from the procedure.

The client language will need to issue a "begin transaction" before calling the procedure, then a commit or rollback depending on the return status of the procedure. In the even of a failure, rollback should be issued.

I'm not a .NET guy so I can't give you actual code syntax, but I guess this should be obvious to your developers.

Cheers

Tim...